Comparison of Genetic Features and Evolution of Global and Chinese Strains of Community-Associated Methicillin-Resistant Staphylococcus aureus ST22

ABSTRACT Methicillin-resistant Staphylococcus aureus (MRSA) sequence type (ST) 22, especially the epidemic MRSA-15 (EMRSA-15), has been one of the most important disease-causing clones transmitting rapidly within and between hospitals globally. However, the genetic features and evolution of Chinese MRSA ST22 remain to be determined. Herein, we performed comparative genomics analysis of 12 ST22 community-associated (CA) MRSA isolates from China with 9 Chinese ST22 CA-MSSA isolates and 284 ST22 genomes from global sources, to clarify the genotypic features and potential transmission of MRSA ST22 strains isolated in China. Phylogenetic reconstruction and time estimation suggested that the Chinese subclade emerged around 2006, and the ST22-SCCmec V clone may have evolved from the native ST22-MSSA clone rather than spread from other regions, indicating that the Chinese ST22-MRSA-V clone is independent of the EMRSA-15 and Gaza clone, with differences in lukSF-PV and tsst-1 carriage. Virulence assays suggested that the ST22-MRSA clone was highly virulent, displaying higher or similar virulence potential as MSSA ST22 predecessors and the epidemic USA300 and ST22-MSSA. However, two nonsense mutations caused by a frameshift in agrC were identified in two ST22-MSSA isolates, resulting in a significant attenuation of virulence. RT-qPCR also demonstrated that the high virulence potential of these ST22 strains may be attributed to elevated expression of agr. This study provides insight into the epidemiology of the novel and highly virulent CA-MRSA ST22 clones. IMPORTANCE Staphylococcus aureus sequence type 22 (ST22) is the main HA-MRSA clone spreading in Europe. It has strong capacity to supplant and replace other formerly epidemic MRSA clones. Previous work has described genotypic characteristics of ST22 belonging to EMRSA-15 and Gaza clone; however, the genetic feature and virulence potential of Chinese spread of ST22 strains are still limited. We conducted a detailed analysis of genomic evolution of global ST22 strains, to clarify the genotypic features and potential transmission of MRSA ST22 strains isolated from China. Our results suggested that the Chinese subclade is highly virulent, and emerged around 2006. We also demonstrated that the ST22-SCCmec V may have evolved from the native ST22-MSSA clone rather than spread from other regions, and the high virulence potential of these ST22 strains may be attributed to the high expression of agr based on the results of virulence assays of Chinese ST22 clones. Our findings are of great importance for providing insights into the epidemiology and pathogenicity of global and Chinese ST22 clones.

constant state of change, with newly adapted emerging MRSA clones predominating in different geographical regions. Furthermore, molecular epidemiology has demonstrated that some MRSA clones are prevalent in geographically limited regions, while others can spread worldwide (1). However, the changes in the prevalence and the epidemiology of different MRSA clones among different geographical regions are still unknown.
Although frequently isolated in many countries, in China, ST22-MRSA is still a sporadic MRSA clone. Our group has isolated a total of 749 MRSA strains from various provinces and cities in China, and detected 12 ST22 CA-MRSA isolates, which caused severe bloodstream infections (12). Recent studies have reported the emergence of the CA-methicillin-susceptible S. aureus (MSSA) ST22-PVL1 clone in China (13)(14)(15); however, detailed information is still limited. Therefore, to describe the genotypic characteristics and possible transmission of CA-MRSA ST22 strains isolated from China, we performed detailed phenotypic and comparative genomic analyses between 12 MRSA ST22 isolates and 9 MSSA ST22 isolates from our database and 284 ST22 isolates from the NCBI.

RESULTS
General characteristics of the global ST22 clones. To characterize the genetic features of ST22 MRSA strains isolated from China, we analyzed publicly available ST22 MRSA genomes (n = 264) from 24 regions and 12 additional ST22 MRSA genomes from our collection, with ST22 MSSA strains used for comparison (20 from the NCBI database and 9 from our collection). The 12 Chinese ST22 MRSA isolates (11 of ST22-SCCmec V and 1 of SCCmec IVh) were isolated from severe CA-MRSA infection cases at different Chinese hospitals, and other global ST22 clones were obtained from Australia, Italy, the United Kingdom, Gaza Strip, and other countries, with widespread geographic distributions (Fig. 1A). Among these ST22 clones, we defined the number of single nucleotide polymorphisms (SNPs) separating isolates to identify possible transmission pairs, together with epidemiological investigation. As shown in Fig. 1B, SNPs of isolates in ST22 clones were from 0 to 452, and the SNPs of gray area in ST22 clones were ,24 SNPs, which indicated that these strains were closely related.
The 305 ST22 genomes were divided into three main clades (Fig. 2). The largest clade, clade A, contained 147 genomes, mostly originating from the United Kingdom, and other regions such as Germany and Singapore. In this clade, all ST22 MRSA isolates harbored SCCmec IVh, belonging to EMRSA-15. Notably, one ST22 MRSA strain (ST22-IVh) from China was found in clade A and gathered with strains from Singapore. Clade B was represented by the Gaza clone, which was mainly isolated from the Gaza Strip. In this clade, almost all ST22 MRSA isolates carried SCCmec IVa. Clade C can be divided into three subclades (CI, CII, and CIII) and two singletons, with clade CII harboring the largest number of genomes (n = 20) from China. Almost all the ST22 MRSA isolates from China (11/12) belonged to SCCmec V, and in the whole phylogenetic tree, the spa type t309 was found only in ST22 clones in China, whereas the ST22-SCCmec IVh clone from China belonged to t032. These results suggested that the "China clone" may be divided into two distinct lineages resulting from two separate acquisitions of the SCCmec elements.
To compare ST22-MRSA isolates in China with those from other regions, we investigated the presence of virulence genes. As shown in Fig. 3, none of the ST22-MRSA strains from clade A contained tsst-1, while almost all clones of clade B (53/54) harbored tsst-1. However, all strains of clade C isolated from China contained PVL (except SKLX100539), while clades A and B did not. In addition to the mobile genetic element (MGE)-induced difference of virulence genes in the genomic architecture of ST22 isolates, most Chinese ST22 isolates (17/21) carried the integrated fibronectin-binding protein (FnBP) locus, while all genomes of clade A and most of B (93.33%, 70/75) had a 2268-bp deletion in the FnBP locus.
To understand the different resistance patterns among the global ST22 strains, we evaluated the presence of resistance genes. There were no significant differences among clades A to C; however, mutations in the grlA and gyrA genes were present in most strains of clade A and clade C excluding Chinese ST22 clones (Fig. 3). It is important to emphasize that the predicted genotypes (resistance to fluoroquinolones) of the ST22-SCCmec IVh clone from China (SKLX111090) were consistent with the result of antimicrobial susceptibility testing (Table 1).
Phylogeographical context and comparative genomics of ST22 strains. Next, we carried out a Bayesian phylogenetic estimation to investigate the global evolutionary history of Chinese ST22-MRSA clone and Gaza clone belonging to non-ST22-A. The Gaza clone BactDating estimated that sraP integrated in clade B ST22 clones was in ;1975. Then in 1994, the vSa4 (containing tsst-1) was integrated in clade BII.
None of the ST22 isolates in clade B contained PVL, while the USa2 (containing PVL) integrated in ST22 clones in clade C emerged in ;1986. Furthermore, our analysis divided clade C into three main clades (I, II, and III) (Fig. 4), and our ST22 isolates were gathered in clade CII that emerged in ;2006. The mapping of ST22 strains showed the independent evolution of ST22 MRSA in clade CII, indicating that the Chinese ST22 lineage (clade CII) was a new subtype that differed from other ST22-SCCmec V clones in different regions. Notably, clade CII clones could be further divided into two subclades (CIIa and CIIb), with clade CIIb2 containing almost all ST22 MRSA strains (10/12) from China. Separation of clade CIIb1 from CIIb2 correlated with the acquisition of SCCmec V. An ST22 MSSA isolate (SKLX57091) was found in CIIb and was more closely related to all MRSA ST22 strains than other MSSA strains in CIIb. Similarly, a MRSA isolate (SKLX100539) found in clade CIIa was surrounded by all MSSA isolates of clade CIIa, indicating that the Chinese MRSA ST22 clones probably arose from ST22-MSSA isolates, independent of the EMRSA-15 and Gaza clone.
Comparison of Staphylococcal genomic islands in ST22 genomes. To understand the differences in the Staphylococcal genomic island and prophages among three clades in ST22 isolates, we analyzed the presence and sequence variations in all ST22 genomes. All ST22 isolates carried a USa2 structure, but only the clade C isolates contained USa2 encoding PVL (Fig. 5A). The sequence alignment showed that, compared with a PVL negative ST22 MRSA strain (SKLX100539) in clade C, an USa2 containing PVL with variable size (from ;6.7-37 kb) was inserted upstream of the riboflavin transporter (RibU) and downstream of the staphylococcal respiratory response AB regulator (SrrAB) in other PVL-positive clade C isolates (Fig. 5A). However, compared with clade A, USa2 in clade B lacked the bcgIAB, which encodes a BcgI-like restriction enzyme.
However, the structure of vSaa in the ST22-MRSA-IVh isolate (SKLX111090) d EMRSA-15 showed the same pattern. Therefore, combined with the results of the phylogenetic analysis, we speculated that the Chinese ST22 MRSA clones that harbored unique SCCmec V may have evolved from methicillin-susceptible ST22 predecessors that acquired resistance to methicillin.
Virulence of Chinese ST22-MRSA isolates in this study. To determine the virulence potential of ST22-MRSA in vivo, mouse models of skin infections and bloodstream infections were constructed. The closely related MSSA strains and the CA-MRSA clone USA300 famous for its hypervirulence, and its Dagr mutant strain (USA300Dagr), were used for comparison. Most ST22-MRSA and ST22-MSSA isolates produced significantly larger lesions than USA300Dagr (P , 0.001), but at levels comparable to USA300 (P . 0.05) ( Fig. 6A and Table S2 in the supplemental material), indicating that both ST22-MRSA and ST22-MSSA were hypervirulent. However, the skin lesions of ST22-MRSA isolates SKLX108615 and SKLX100539 were much larger than that of the closely related ST22-MSSA clones SKLX57091 and SKLX56567 (P , 0.001), respectively. In bloodstream infection models, as shown in Fig. 6B and Table S3, mice infected with ST22-MRSA strains and ST22-MSSA strains (except SKLX57091 and SKLX56567) all died within 72 h (n = 10), while those infected with the closely related ST22-MSSA strains SKLX57091 and SKLX56567 were still alive until euthanasia. To further investigate the virulence of ST22-MSSA strains SKLX57091 and SKLX56567, we then compared the alpha-toxin activity by analyzing lysis of rabbit RBCs. Alpha-toxin activity often represents agr expression in strains. As shown in Fig. 6C and Fig. S1, we found that all ST22 strains displayed much higher hemolytic capacity than ST22-MSSA clones SKLX57091 and SKLX56567 (P , 0.01).
agr dysfunction accounts for the low virulence of ST22-MSSA. We next compared the sequence of agr in ST22-MRSA and ST22-MSSA. In ST22-MSSA SKLX57091 and SKLX56567, two nonsense mutations caused by frameshift in agrC (c.349delT and c.1140insT, respectively) were identified (Fig. S2A). These mutations accounted for the defect in agr expression and the avirulence of the strains. We next performed RT-qPCR to evaluate the expression of hla and agrC, which regulate toxins such as alpha-toxin and PSMa that are most frequently associated with CA-MRSA virulence among ST22-MRSA and ST22-MSSA isolates. All hla and agrC genes were highly expressed in ST22  Strains  ERY  CLI  OXA  Pen  SXT  TCY  VAN  RIF  CIP  LVX  MFX  GEN  AMK  TGC  LNZ  isolates except SKLX57091 and SKLX56567 (P , 0.01, Fig. S2B). Thus, the agr dysfunction may be responsible for the low virulence of the SKLX57091 and SKLX56567.

DISCUSSION
MRSA ST22, especially the EMRSA-15, has spread across most countries, especially in Australia and the United Kingdom (4, [16][17][18]. However, as a sporadically occurring MRSA clone in China, the genomic features and clonal origin of ST22-MRSA have not been fully identified. Therefore, herein, we characterized 12 ST22-MRSA isolates, which cause life-threatening infections. We separated the ST22-MRSA isolates from China as two clones: t309-carrying SCCmec V clone and t032-carrying SCCmec IVh clone. The ST22-SCCmec V clone was  Our results suggested that the "Chinese clone" consists of two distinct lineages resulting from two separate acquisitions of the SCCmec cassettes. The acquisition of SCCmec V cassettes was a critical genetic event in the evolution of the Chinese ST22-MRSA clone resistant to b-lactam antibiotics. Notably, compared with MSSA strains, the larger SCCmec cassettes in MRSA strains have been reported to reduce the bacterial toxicity due to the expression of mecA, which can repress the expression of cytolytic toxins (19). Collins et al. (20) reported that this significant toxin repression could compensate for the energy-consuming maintenance of a large SCCmec element and its associated resistance to methicillin, but probably leads to reduced fitness outside of hospital settings. However, the virulence assays suggested that ST22-MRSA shows a level of virulence comparable to that of ST22-MSSA. mecA can inhibit agr expression; however, He et al. (21) have reported that the evolution of CA-MRSA clones is always accompanied by a very low methicillin resistance level. This may be associated with the reservation of the high virulence level of the MSSA predecessors of CA-MRSA. In this study, the MIC of oxacillin in ST22 is 4 mg/L, which is a breakpoint that served to classify MRSA. Therefore, our results highlight that the evolution of ST22 MSSA into ST22 MRSA was accompanied by low methicillin resistance level.
S. aureus produces many secreted virulence molecules, including the most prominent pore-forming toxins Hla, PSMa, and PVL, known to contribute to skin and soft infections, sepsis, and pneumonia (22). The production of these toxins is strictly controlled by global virulence regulatory system agr, and the occurrence of mutations in agr affect the ability of S. aureus to cause human infections (23,24). In the present study, we identified the ST22-MRSA clone and its closely related ST22-MSSA from China, which have the same genetic features (ST22, t309, and identical virulence and resistance profile), displaying completely different virulence phenotypic response. The ST22-MRSA isolates were hypervirulent while the ST22-MSSA showed a very low level of virulence. Of note, two nonsense mutations caused by frameshift in agrC (c.349delT and c.1140insT, respectively) were observed in two ST22-MSSA isolates, which resulted in a significant attenuation of virulence. Thus, our study emphasizes the great importance of agr expression in maintaining the high toxicity of MSSA predecessors during CA-MRSA evolution.
In conclusion, this study provided the first detailed characterization of ST22-CA-MRSA strains. We determined that the ST22-MRSA-V clones from China are independent of the EMRSA-15 and Gaza clones. We also investigated the evolutionary history of a new and highly virulent ST22-MRSA clone, and revealed it belonged to sporadic MRSA clones that are rare and not epidemic in China. Nonetheless, sporadic MRSA isolates may constitute a significant extensive reservoir of virulence genes located on MGEs, and our clinical and experimental data indicated that ST22-MRSA is hypervirulent and has a strong ability to cause life-threatening diseases. Thus, the transmission of ST22-MRSA should be of concern. Furthermore, our findings are of great importance for elucidating the role of agr expression in CA-MRSA evolution. A major limitation of this study is that, due to a lack of available clinical records, we only identified 12 ST22 MRSA isolates as CA-MRSA in accordance with the Centers for Disease Control and Prevention definition among 749 MRSA strains from China between 2014 and 2019. This means that such a small sample size of the ST22-MRSA clone may result in the difficulty of evaluating the transmission tendency and the potential threat of this clone in China, both outside and within health care settings. Therefore, the findings about the ST22 clone from China should be interpreted with caution. Further studies will be required to clarify how the ST22 MSSA isolates from China evolved to obtain methicillin resistance and the virulence of ST22 MRSA isolates and ST22 MSSA isolates. Furthermore, we will also record as many clinical data as possible in the future data collection to distinguish community infection from hospital infection.

MATERIALS AND METHODS
Collection of bacterial isolates. Thirty-three ST22 MRSA and 68 MSSA ST22 isolates were collected from patients with bloodstream infections in 18 provinces in China from 2014 to 2019. Among these ST22 S. aureus isolates, 21 were classified as CA according to the Centers for Disease Control and Prevention definition (12 CA-MRSA and 9 CA-MSSA). The antimicrobial susceptibilities of 21 ST22 isolates to 16 antimicrobial agents were determined using the agar dilution method in accordance with the Clinical and Laboratory Standards Institute protocols (25).
Whole genome sequencing and genomic analysis of global ST22 clones. Whole genome sequencing of the ST22 genomes was carried out on a HiSeq X 10-PE150 platform (Illumina, San Diego, CA, USA). SPAdes v3.14.1 (26) was used to assemble the processed Illumina data (after adaptor trimming and quality filtering). The genome annotation of ST22 isolates was performed with DFAST-core v1.2.11 (27). Genomes of additional 284 ST22 strains from other regions were downloaded from GenBank for comparison (Table S1).
Comparison of antimicrobial resistance genes, virulence factors, pathogenicity islands, and prophages among ST22 strains. Antimicrobial resistance genes and virulence factors were identified by ABRicate v1.0.0 (https://github.com/tseemann/abricate) based on CARD and VFDB databases with default settings. Pathogenicity islands and prophages were compared using BLAST, with sequences of those of S. aureus as queries.
Virulence assays. All animal experiments were approved by the Institutional Animal Care and Ethics Committee of the First Affiliated Hospital of Zhejiang University, School of Medicine. To evaluate the virulence potential of ST22-MRSA in vivo, mouse skin infection and the bloodstream infection models were established as previously described (31). Briefly, for the former, 6-week-old BALB/c-nude female mice were used for this study. Mice were inoculated with 100 mL sterile NaCl solution containing OD 600 = 0.5 live bacterial cells by subcutaneous injection (each group injected 5 mice). Abscess length (L) and width (W) values were measured to calculate the size of skin abscesses, i.e., abscess size = L Â W. We measured and recorded lesions on the skin of each mouse 2 days after injection using calipers, and the normal blank group of mice injected with NaCl solution alone was used as the negative control. For the latter, 6week-old BALB/c female mice were used. Mice were inoculated with 100 mL sterile NaCl solution containing 2 Â 10 8 live S. aureus by caudal vein injection (each group injected 10 mice). The normal blank group of mice injected with NaCl solution alone was used as negative control. After inoculation, the physical condition of each mouse was monitored and recorded every day. When the first mouse infected with S. aureus strains died, we euthanized all remaining mice for each group. As hemolysis (lysis of erythrocytes) is a crucial determinant of S. aureus virulence, we accessed hemolytic capacity of S. aureus strains as previously described (31). ST22 bacterial cultures were cultivated in TSB for 9 h and centrifuged. The supernatant was incubated with rabbit red blood cells (RBCs) for 3 h at 37°C. Hemolytic capacity was determined by measuring the optical density at 540 nm using a Micro ELISA Autoreader. Purified water was used as positive control, and rabbit RBCs resuspended in NaCl solution were used as negative control.
Quantitative reverse-transcription (qRT)-PCR. The expression of hla and agrC in ST22 isolates was evaluated by qRT-PCR, with gyrB as an internal control. We chose S. aureus USA300_FPR3757 (ATCC BAA-1556) as a reference strain due to its high expression of these genes (32).
Statistical analysis. All assays were performed in triplicate independently. The difference in skin lesions between mice was examined using q test. Kaplan-Meier analysis was performed to evaluate differences in the survival rates of mice infected with S. aureus. A chi-squared test was used to analyze the results of the hemolytic test. Unpaired t test was used to analyze the qRT-PCR data. All data were analyzed using GraphPad Prism 9.0 software (GraphPad, San Diego, CA, USA). P , 0.05 was considered as statistically significant.
Data availability. The whole-genome sequences of 21 ST22 S. aureus isolates from this study have been deposited in the GenBank database under BioProject accession no. PRJNA769995.